You are here
SmartWatch Gestures Dataset
The SmartWatch Gestures Dataset has been collected to evaluate several gesture recognition algorithms for interacting with mobile applications using arm gestures.
Eight different users performed twenty repetitions of twenty different gestures, for a total of 3200 sequences. Each sequence contains acceleration data from the 3-axis accelerometer of a first generation Sony SmartWatch™, as well as timestamps from the different clock sources available on an Android device. The smartwatch was worn on the user's right wrist. The gestures have been manually segmented by the users performing them by tapping the smartwatch screen at the beginning and at the end of every repetition.
Available for download. This gesture archive is provided for research or academic purposes only. Publications that include results obtained with this dataset should please refer to the following publication:
Gabriele Costante, Lorenzo Porzi, Oswald Lanz, Paolo Valigi, Elisa Ricci, Personalizing a Smartwatch-based Gesture Interface With Transfer Learning, 22nd European Signal Processing Conference (EUSIPCO), 2014